A parameterized runtime analysis of evolutionary algorithms for MAX-2-SAT

  • Authors:

  • Andrew M. Sutton;Jareth Day;Frank Neumann

  • Affiliations:

  • University of Adelaide, Adelaide, Australia;University of Adelaide, Adelaide, Australia;University of Adelaide, Adelaide, Australia

  • Venue:
  • Proceedings of the 14th annual conference on Genetic and evolutionary computation
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

We investigate the MAX-2-SAT problem and study evolutionary algorithms by parameterized runtime analysis. The parameterized runtime analysis of evolutionary algorithms has been initiated recently and reveals new insights into which type of instances of NP-hard combinatorial optimization problems are hard to solve by evolutionary computing methods. We show that a variant of the (1+1) EA is a fixed-parameter evolutionary algorithm with respect to the standard parameterization for MAX-2-SAT. Furthermore, we study how the dependencies between the variables affect problem difficulty and present fixed-parameter evolutionary algorithms for the MAX-(2,3)-SAT problem where the studied parameter is the diameter of the variable graph.